Virtualisation and cloud technologies are storming their way into enterprise infrastructure deployments. But, what happens to networks and their management, with the advent of these technologies? Sathya Mithra Ashok gets some answers.
Network management is no longer what it used to be. The advent of increasing virtualisation and the spectre of cloud computing has to an extent changed what an organisation wants from a network, and how it can go about achieving its requirements.
“According to Gartner, through 2013, at least 60% of enterprises will experience slow or inconsistent application performance due to improper network design. Regardless of the cloud model — private, public or hybrid — the simplicity, uptime and capabilities of the network will ultimately define application performance and the user experience in the
cloud, and networks must be cloud-optimised at every critical point,” says Ali Ahmar, regional sales manager MENA at Brocade.
“For cloud computing solutions to succeed, compatibility with existing networks is key. Without this, the technology is dead on arrival. But, if we peel the cloud computing onion just one layer, its obvious that networking will need to evolve to keep pace. In a cloud, new sources and destinations appear and disappear at a moment’s notice, various organisations share the same multitenant infrastructure. And that’s just the server estate. Today’s network has been optimised for a static, host-centric, presegmented network. Clouds are dynamic and multitenant. It’s a good bet that current network topologies will need to be rewired,” says Deepak Narain, manager, systems engineering MENA for VMWare.
“The network needs to be designed and built to ensure flexibility, scalability and high availability. From a management perspective, it will ask for easy allocation of devices and bandwidth, proactive identification of faults and security threats, and managing performance to meet the response time of applications,” says Manish Mishra, Middle East VP for HCL Technologies.
“Cloud infrastructures inherently push users to create traffic over WAN, which creates two problems. First, both business critical (such as Salesforce.com) and productivitydraining (such as YouTube) traffic must contend for the same bandwidth. To solve this, IT organisations need a way to manage the allocation of bandwidth by application,” says Florian Malecki, EMEA senior product marketing manager at SonicWall.
Building the network
With almost every enterprise in the Middle East seriously considering private or hybrid clouds in the near future, preparing networks for what is to come becomes of utmost importance. And that involves two different activities – one, building the network right and the second, managing the network right.
“Every technology brings benefits coupled with challenges. Though cloud technology brings lot of benefits, organisations need to develop strategies on how to meet the challenges around deploying the right network environment. Enterprises should decide which applications should go on cloud based on business requirements and security considerations, and pinpoint the specific network needs of these applications.
They have to re-architect the existing network environment for seamless access of applications and ensure a multi-level security architecture for data privacy. They also have to monitor performance of applications from each location and how they are handled on the cloud, have real-time load assessment and traffic redirection based on the performance and availability of applications and deploy flexible resource options with a “pay-as-youuse” option over a certain minimum required bandwidth,” says HCL’s Mishra.
Nicolai Solling, director of technology at HelpAG Middle East says, “One of the key technical factors in focusing on cloud computing is to understand the apps operating on your network making sure that cloud based services are getting the correct service level and priority in your network. Any services considered for the cloud should always be evaluated and verified if such inhibitors such as limited bandwidth, high latency or packet loss will make the solution unusable from a time perspective. For instance, if a company is considering moving data backup to the cloud, the backup aspects may be completely acceptable, but what will the implications be if large systems need to be recreated from the cloud? Potentially in order to recreate systems you may need to move multi-gigabits of traffic which can take too long compared to restoring from a local data source such as disc or a tape drive.”
“To fully capitalise on the promise of the cloud, CIOs need to build integrated solutions – from the cloud lifecycle management, all the way down to the server, storage, LAN and WAN management tools. This will provide a single cloud management infrastructure that will offer both end-to-end first-alert visibility and end-to-end service provisioning,” adds Mashood Ahmed, regional managing director for Ciena in the Middle East.
Managing the network
While the first part of winning the battle with a strong network for cloud technologies is choosing right and implementing it, the second (albeit harder part of the battle) is managing and monitoring the network in an efficient fashion.
“Today’s networks are highly complex in terms of service richness and flexibility, which presents new challenges to the network management operator. However, the network devices now also offer much greater intelligence and agility – when harnessed with the correct tools they allow more control and efficient use of the network resources. Service planning and provisioning systems must provide ease of use and the flexibility to optimise bandwidth, which is critical to ensuring capacity can be deployed cost efficiently. These systems can further maximise operational savings by taking advantage of a software based multilayer mesh control plane to automatically compute service paths according to the operator’s required constraints. In addition to traditional network protection mechanisms, the agile, intelligent network allows for a blend of service restoration schemes to meet customer SLAs, while maximising the use of network infrastructure,” says Ciena’s Ahmed.
“As the cloud evolves, network management will have to evolve to not only depict usage and utilisation, but also to be able to allocate and rescind network resources. This function will evolve towards not only maintaining a status check on devices, but also monitoring the real time allocation of network resources vis-à-vis the business needs. Traditional administrative tools will not be discarded but will need to adapt to measuring and monitoring availability and usage to deliver consistency across different network components for a seamless application offering to the cloud user,” says HCL’s Mishra.
Solling adds, “In general, any aspect of network management involves a greater understanding of the requirements based on the network equipment and the data flowing on it. Therefore, a decent understanding of the network and data requirements will allow managers and CIOs to make the correct decision for scaling the infrastructure. Quite often equipment is oversold or over-specified because there is little understanding of what kind of data the network equipment will have to carry.”
Overselling of equipment is not the only challenge that network and IT managers have to deal with in the Middle East.
“Managing bandwidth for cloud applications, and users bandwidth efficiency and management is paramount when
consuming business applications from the cloud, particularly for those applications that are driven by massive databases served from multiple virtual environments. However, the on-premise bandwidth used to access the cloud is typically limited. Organisations need the capability to prioritise bandwidth allocation in addition to user access for more important business applications over less important ones,” states SonicWall’s Malecki.
Mishra states, “In our experience, challenges for network management are universal, but the level varies by region. In the Middle East, we have seen that adopting and maintaining process, and acquiring and retaining skills both for legacy and modern technologies, are the prime challenges.”
“The Middle East is a very diverse region, and so a number of factors do in fact apply here. Large geographies, for instance, and high incidence of submarine cable systems lead to frequent fibre outages. A network management organisation relies on the necessary network agility and flexible tools to rapidly restore service and troubleshoot network issues,” says Ahmed.
Apart from the purely technical challenges, there are other issues that managers face on a regular basis in connection
to the teams that work on the projects.
As Solling points out, “Looking at management solutions there needs to be a flow of data from different devices in the
network quite often managed by different departments. Sometimes it can be difficult to get full support for getting the data into the system. As an example, network management solutions may be managed by the network team but the server team may find it very irrelevant to do the needful to support the solution. Typically an explanation of why this is important addresses this and even the server team realises that a network management solution is a great tool for them too. Another challenge is that network management solutions are typically handled as a sub-component of a larger project, when in fact the importance of network management actually justifies this to be a separate project.”
Challenges or not, the cloud is inevitable — whether private, hybrid or public — for most enterprises in the Middle East. And with that, organisations will learn to overcome or solve the issues connected with networks and their management that they might be facing or are likely to face. This will most likely have to be done despite service providers’ restrictive abilities in the short term. A growing demand for better networks is likely to ensure better service provision from ISPs in the future.
Network management solutions themselves are also most likely to evolve in the coming years to suit the changing requirements of enterprises.
“All aspects of the cloud will be managed in an orchestrated and integrated fashion. Traditional technologies will be subsumed and abstracted into cloud services. In the network’s case, bandwidth services will be represented as higher level cloud services. Rather than acquiring a 10GE connectivity service, cloud solutions will deliver 100TB of managed capacity offering resilient, location independent storage. The same would be the case, for instance, with compute capacity. In order to deal with these new high level services, management solutions must bind together all of the cooperating and dependent technologies into a single, cohesive, end to end cloud management solution,” states Ahmed.
“As more cloud services get deployed, network management should shift from monitoring infrastructure to managing service availability and performance. Monitoring and management solutions of network devices, circuits and locations
will need to relate traditional measures like device level thresholds to end-to-end application response time. For example, high CPU in a core switch at a data centre will need to alert the cloud administrator that a specific application could potentially be affected and also provide a solution like traffic re-direction at the click of a button in real-time. The tools that dynamically assign compute and storage resources to application requirements will have to encompass network resources as well, and these will need to be tied to the monitoring solutions for real-time updates,” adds Solling.
Whether these evolutions meet changing requirements, and whether the service providers market changes enough to accommodate the dynamic needs of Midlde East enterprises, are matters that will be decided in time.